Title:Bregman Divergence-Based Data Integration with Application to Polygenic Risk Score (PRS) Heterogeneity Adjustment

Abstract:Polygenic risk scores (PRS) have recently received much attention for genetics risk prediction. While successful for the Caucasian population, the PRS based on the minority population suffer from small sample sizes, high dimensionality and low signal-to-noise ratios, exacerbating already severe health disparities. Due to population heterogeneity, direct trans-ethnic prediction by utilizing the Caucasian model for the minority population also has limited performance. In addition, due to data privacy, the individual genotype data is not accessible for either the Caucasian population or the minority population. To address these challenges, we propose a Bregman divergence-based estimation procedure to measure and optimally balance the information from different populations. The proposed method only requires the use of encrypted summary statistics and improves the PRS performance for ethnic minority groups by incorporating additional information. We provide the asymptotic consistency and weak oracle property for the proposed method. Simulations and real data analyses also show its advantages in prediction and variable selection.